Each note takes a single planning problem, works an example through with real numbers, and says where the method stops being useful. Written by the team that builds the platform.
What changes under CSRD is that supply chain figures must be reproducible by someone else a year later. The datapoints a planning system can own, and the ones that belong somewhere else.
Wave planning decides when work reaches the floor. Working release times back from carrier cut-offs, sizing a wave against the packing drain rate, and when waveless release fits better.
Tanker scheduling disruption as a priced decision: which commitments survive a chokepoint closure, what the alternatives cost, and how to prepare in advance.
Tanker scheduling disruption as a priced decision: which commitments survive a chokepoint closure, what the alternatives cost, and how to prepare in advance.
How methane emissions measurement reporting verification reconciles factor based inventories with measured data, and why detection to repair is the metric.
Energy demand forecasting where joint production, term contracts and no inventory buffer break standard method. What transfers, and where scenarios take over.
Localization portfolio prioritization as a constrained selection problem: what to maximise, the inputs that decide the answer, and how sequencing changes it.
Terminal and berth scheduling on the queueing curve: why the last few points of utilisation cost so much, demurrage as a partial view, and picking a target.
Commodity hedging exposure has to be measured before a ratio means anything. The exposure inventory, where unintended positions live, and basis risk you keep.
LNG supply planning as an inventory problem at both ends: what a slipped slot costs in days of tank, diversion economics, and regasification slot rules.
Petrochemical margin optimisation when co-products carry a third of revenue: the three quantities called margin, feed choice, severity and transfer price.
Fuel retail supply one compartment at a time: the tank with a random drain, what a drop costs per litre, fixed delivery days, and wet stock reconciliation.
Bunker fuel procurement as three decisions in one purchase: the greedy lifting rule, the cost of deviating for a cheaper port, and ISO 8217 as a constraint.
Power demand forecasting across four horizons: the temperature response, why scoring with weather known flatters the number, and settlement arithmetic.
Renewable generation forecasting scored on averages and sold into markets pricing extremes: the power curve, ramps, aggregation and probabilistic output.
Battery storage dispatch with the cycle cost priced in: the spread that has to clear, why the revenue stack competes with itself, and hindsight backtests.
What a place in the grid connection queue is worth: Little's Law on the wait, why one withdrawal stalls everyone behind it, and the cost of delay.
Oilfield materials planning against a schedule that moves monthly: where newsvendor logic justifies the stock, Palm's theorem on spares, remote surplus.
Drilling rig scheduling as an asymmetric travelling salesman problem where the move cost dominates, plus skewed durations and what a rig contract buys.
Scope growth in turnaround planning: the cost of a day, three sources of scope, why schedules built on expected durations finish late, and discovery.
Refinery scheduling against a monthly plan built in averages: the tank coupling the plan cannot see, the pooling problem, and giveaway priced per day.
Local content economic impact comes from the Leontief inverse rather than the headline percentage. Direct, indirect and induced effects, and what it assumes.
Flaring reduction economics at a site with no tie-in: routine versus safety volume, the options ranked by build cost, and valuing what you actually recover.
Asset integrity predictive maintenance past a code interval: what each step costs, rotating against static equipment, and the turnaround scope it feeds.
Process mining industrial operations against the event log: where it earns its keep, findings that recur, conformance checking, and what a deviation means.
Contract intelligence energy operators can act on: the obligation register, why extraction is hard here, provenance as a requirement, and human review.
Supplier onboarding at scale as a throughput problem: where the queue forms, what Little's Law says about pipeline, risk tiering, and measuring the funnel.
Workforce localization planning as a stock and flow problem: why quota hiring buys attrition, which roles are constrained, and retention at eighteen months.
Crude differentials pricing beyond last month plus an adjustment: what the differential compensates for, regional netbacks, and the slow feedback loop.
Drum buffer rope governs release from the constraint and reads buffer penetration daily. This covers the release arithmetic, the protective capacity it assumes, and what changes when the bottleneck moves.
Why chained substitution rules have to terminate, what a downgrade really costs once the premium item is tight, and the three fields that stop a substitution corrupting both forecasts.
How to build a rationing rule from facts of record, defend it to a buyer without publishing the weights, and stop the shortage from rewriting the demand history it rationed.
Decoupling points, buffer zone arithmetic and the lead time claim, tested against a well parameterised MRP carrying the same stock, so you can tell which part of the gain is real.
Why a source plant sees one 350 unit week and four empty ones, how the same arithmetic run at every shipping point makes that demand knowable, and when reorder points win.
How forecast consumption and the demand time fence turn a demand plan into a buildable statement, the two ways that arithmetic goes wrong, and what the schedule owes order management.
Why an MRP run rewrites orders that were fine last week, how lot sizing creates the lumpiness it was meant to absorb, and which dampener to reach for first.
Forecast consumption decides how orders net against the demand plan inside the execution window. How the consumption windows misfire, where double counting hides, and what to reconcile weekly.
Why route optimization loses to the routes drivers already know: what the solver solves, what master routes buy, adherence, and unmeasured service times.
Sales territory design drifts after the project ends. What balance means, the compactness trade, the price of moving an account, and reviewing on a cadence.
Fleet sizing against a peak that lasts four weeks a year: asymmetric error costs, variability over volume, own against contract or spot, and the mix decision.
Pricing the delivery time windows you already give away: what window width costs, promise against plan, letting customers choose, and measuring arrival times.
Warehouse slotting past velocity ranking: the cube per order index, order affinity, the golden zone against lifting limits, and what a move list costs.
Truck load building as bin packing with side conditions: the freight density that decides whether weight or cube binds, and how to measure fill honestly.
A supply chain control tower that ends at a coloured icon has automated two stages of four. What an exception queue needs, and why the loop has to write back.
Cold chain management as an inventory problem: writing an excursion to the batch status planners read, mean kinetic temperature, and FEFO with a tolerance.
Freight procurement judged after the tender rather than at award: what an unenforceable rate is worth, the split award case, and routing guide compliance.
3PL capacity planning after the warehouse is outsourced: the volume band on the invoice, forecasting in the provider's units, and what priority costs at peak.
A Kalman filter tracks demand through missing weeks, level shifts and disagreeing data sources. What the two variance parameters control, how to estimate them, and how to read the innovations.
Rolling origin design decides which forecasting model wins your bake-off. How many origins you need, when to refit, how to pool errors across items, and which comparisons are safe.
Censored demand forecasting recovers the sales a stockout hid, why the loop keeps shrinking the buffer, four recovery methods, and the flags they need.
Forecastability measured before you buy: the naive baseline floor, how error explodes as the grain gets finer, and what an accuracy quote without one hides.
Sell-in vs sell-out forecasting: the distortions inside shipment history, building the inventory bridge to consumption, and working with partial visibility.
Promotion uplift modelling separates incremental volume from pull-forward and cannibalisation, starting with the baseline and the counterfactual under it.
Intermittent demand forecasting for the part of a catalogue that sells rarely: classify first, Croston and its successors, and why the usual metrics fail.
Forecast value add scored stage by stage against actuals: how to read a negative stage, the four measurement mistakes, and what changes once people see it.
Hierarchical forecast reconciliation when regional and national numbers disagree: what bottom-up and top-down discard, and what minimum trace uses instead.
Two tests separate probabilistic demand forecasting from a wider confidence band: whether lead time is random, and whether the output carries its economics.
Forecasting model selection by pedigree: which data shapes each model family was built for, four questions that narrow the list, and where ensembles cost you.
Demand sensing is a latency upgrade rather than an accuracy one. The five mechanics, where the value lands, and the horizon at which the correction decays.
Price elasticity by pack and channel: where the spread comes from, the bias in the standard log-log regression, and why cross-price often dominates it.
Tariff cost pass-through pricing decided per item and channel, what has to be true of landed cost first, and why the supply consequence belongs in it.
Trade spend allocation as a reallocation problem: moving budget toward measured returns, the patterns that recur, and the supply check before it is set.
Price pack architecture is a design decision before it is an analysis. How to space the rungs, price the gaps, and stop the middle pack absorbing the ladder.
Rebate management fails on structure before administration: why a retroactive tier creates a cliff, what the accrual costs, and how to price the increment.
Channel conflict pricing starts with arithmetic: when the gap between two of your own net prices beats the cost of moving goods between them, the leak opens.
The account plan and the item forecast cross rather than nest, so nothing forces them to agree. The grain that does, and what one thirty percent customer adds to the buffer.
Stacked tier margins compound, so a distributor margin structure has to be built backwards from the shelf. Five functions sit inside the percentage and only two scale with your price.
Under a binding supply constraint, price rations at no volume cost, so the number to estimate is the market clearing price. Weigh the contribution against the customers who qualify a second supplier.
Planning system testing in three layers: exact arithmetic identities, properties that hold whatever the numbers are, and comparison to a known baseline.
An SAP IBP implementation turns on decisions made early: planning area structure, the time profile, key figure disaggregation and how order data arrives.
Oracle supply chain planning cloud deployments turn on collections design: what gets collected, how often, and what a back dated correction does to it.
An Excel planning migration succeeds when it is sequenced by value: what to keep, what to drop, the reconciliation contract and the workbook owner.
ERP vs planning system boundaries come down to who owns each field. The four contested ones, and how to specify the write back and the override policy.
A planning data warehouse earns its keep in the semantic layer: conformance decisions, versioned hierarchies and measure definitions written in sentences.
Choose planning system integration patterns by what a source restatement does to each, then check idempotency, replay and the outbound direction.
Demand planning software selection fails on long requirement lists. The twenty areas where products differ, plus demo scripts and reference questions.
ABC XYZ analysis that changes a policy rather than producing a deck: what hangs off each of the nine cells, where to store it, and when to re-segment.
Planning exception management when the queue is longer than the hour available: setting thresholds by consequence, grouping causes, ageing, and clearance.
Replenishment parameters set at go-live and never revisited: which ones exist, who may change them, trigger based review, and the audit that finds the worst.
Touchless planning starts by measuring which lines a human touch improves. Finding the segment, the guardrails, a staged rollout, and what to watch once on.
Demand planner workload has no benchmark. Where the time goes when measured, span of control as an equation, touch rate, and what to take off the desk first.
Supply chain master data governance for the fields that break planning: ownership per field, validation at creation, standing checks, and the defect position.
Standing up planning for a business being sold, when the item master has one hierarchy, a shared plant serves both owners, and the transitional service agreement has an end date.
S&OP vs IBP in operational terms: a financial reconciliation that closes, the gap as an owned object, scenarios before the meeting, named assumptions.
A consensus forecasting process works when the room argues about assumptions rather than the number, and when padding is measured before it is managed.
Planning assumptions management with a scored ledger: what belongs in it, how to score entries next cycle, and how to keep it small enough to survive.
Supply chain scenario planning that moves cash: overlay scenarios instead of copies, financial measures declared in the model, committing the winner.
Why collaborative planning forecasting replenishment stalls after the data exchange, the four commitments a working arrangement needs, and how to test yours.
Vendor managed inventory moves the replenishment decision to the supplier. What has to be agreed, why your own stock can rise, and the data it depends on.
Joint business planning fails when commercial teams commit to volumes supply never saw. How to separate commitments from ambitions and sequence the cycle.
Supplier collaboration works once the forecast you send carries a commitment. How to separate forecast, plan and reservation, and what should come back.
Distributor reporting to principal planners fails on late, unmapped files. The fields needed, the inventory identity check, and what the principal owes back.
Agricultural supply planning after harvest: intake capacity, bin segregation, drying limits and the carry decision that fixes what the year can sell.
Crop yield forecasting improves in steps through a season. How area estimation, method combination and asymmetric decisions change the number you use.
Commodity procurement timing set by how long your selling price is fixed: separating the price decision from the physical buy, and scoring the buyer.
Perishable supply chain planning treats shelf life as a budget spent from harvest, with thermal history, ripening control and lot allocation as levers.
Seed and planting planning across a three year multiplication cycle: contracted area, germination thresholds, carryover risk and variety life cycles.
Six buyer questions that separate an agentic AI supply chain product from relabelled automation, covering guardrails, audit records and measurement.
SAP APO end of life alternatives set against the 31 December 2027 maintenance date, the four viable paths, and the custom logic count that decides them.
A supply chain planning proof of value protocol: the history to supply, rolling origin folds, the two baselines to require, and how results get flattered.
Supply chain planning implementation failure is quiet: a planner finds three wrong recommendations and the work moves back to a spreadsheet. What to check.
Multi-echelon inventory optimization stops every node buffering the same variability: guaranteed service, stochastic service, and the inputs that decide it.
Four assumptions inside the standard safety stock calculation break in ordinary conditions, each in a direction that costs money. Which one to fix first.
Lead time variability drives more inventory than demand noise: demand over lead time, why an ocean right tail breaks normality, and the order date.
Inventory rebalancing checks the network sideways before a purchase order is raised: consolidating transfer candidates, the obstacles, and when to buy.
Supply chain network design as capacitated facility location: the costs that never enter the model, service as a constraint, and the transition it ignores.
Greenfield analysis deletes the estate and asks where sites would sit on the demand map: the centre of gravity, choosing k, and the cost of your own history.
A supply chain digital twin adds the weekly clock an optimiser leaves out: deliberate disruption runs, recorded seeds, and honest replication counts.
A postponement strategy moves the decoupling point later so variants pool: where the arithmetic holds, the three forms, and why product design decides it.
What planning controls inside the cash conversion cycle: turning a week of coverage into a number finance uses, where to measure it, and the payables trap.
Currency timing purchasing decisions with the forward curve rather than a house view: what trades off, why buying early usually fails, and hedging instead.
Inventory working capital targets that land where nobody chose: why a uniform percentage misallocates, the service frontier, and four kinds of stock.
A cost to serve analysis allocated as a percentage of revenue can only reproduce the gross margin ranking, so the drivers have to be counts somebody measured.
Sales and operations execution as a decision rights problem: what the weekly forum may change inside the time fence, and what it has to escalate instead.
Warehouse labour planning converts a demand plan into hours using lines rather than units, an intraday profile, and a flexible layer ranked by notice period.
Returns forecasting as a convolution of sales and a lag distribution fitted from your own history, split by cause, with grading modelled as a lead time.
Service parts planning driven by the installed base rather than the part's own history: the two lifecycle turns, the all-time buy, and fill rate at the van.
Sequencing a move off an incumbent so supply holds through the crossover, covering tooling title and transferability, the qualification calendar, the bridge buy arithmetic and the wind-down clauses that decide the cost.
Sub-tier supplier risk hides behind healthy tier one scorecards. Building the map without full disclosure, entity resolution, and acting on a concentration.
Supply chain risk prioritisation by consequence rather than event severity: simulating criticality first, three refinements, and making an alert actionable.
Wave planning decides when work reaches the floor. Working release times back from carrier cut-offs, sizing a wave against the packing drain rate, and when waveless release fits better.
Travel dominates manual picking time, yet most sites walk a location sort nobody selected. How pick density picks the routing rule, why batching pays more, and what to measure.
Demurrage and detention are generated by appointment capacity, chassis supply and receiving hours. How to forecast the charge, budget its tail, and place the decision where someone can act.
Rate levels are close to unforecastable at long horizons while vessel supply is largely known in advance. What to forecast, how to score it honestly, and which decisions it should change.
Why the product hierarchy inherited from finance or merchandising rarely suits planning, which attributes to store instead of fixed levels, and how to choose the level you forecast at.
How unit of measure conversion errors enter a planning run, why a wrong factor hides from every variance check, and the receipt-based test that finds them in an afternoon.
Why the headline discount tier is often the worst commitment to sign, how to price a take or pay shortfall before agreeing it, and which clauses change the arithmetic.
How to read the incoterm on a purchase order as a planning parameter: which lead time legs you own, when in-transit stock becomes yours, and where cargo risk sits.
Most tail spend is too small to source competitively, and the arithmetic proves it. Where the break-even sits, which mechanisms move it, and why spend is the wrong ranking variable.
Standard cost includes overhead that stays behind when the volume goes, so most make or buy analysis compares the wrong two numbers. How to rebuild it on avoidable cost and break-even volume.
Finite capacity planning against an MRP run that assumes infinite capacity: what a constrained solve adds, one shared constraint set, and the data to fix.
Available to promise as a netting calculation against the master schedule: what to do when supply is short, three allocation policies, and re-promising.
A 45-MINUTE WALKTHROUGH ON YOUR OWN DATA. NO SLIDEWARE.